Steven Joffe and Holly Fernandez Lynch teach at the University of Pennsylvania Perelman School of Medicine. Joffe is a professor and interim chair of the Department of Medical Ethics and Health Policy, and a professor of pediatrics. Fernandez Lynch is an assistant professor of medical ethics in the department.

Vaccines are widely expected to offer the best way out of the covid-19 pandemic. But they won’t help unless they’re safe and effective, making it essential that only rigorously tested approaches reach the public. Recent history suggests there is reason to be concerned.

On May 18, the biotechnology company Moderna reported preliminary results from its first-in-human coronavirus vaccine trial. Releasing limited information — though to great fanfare — Moderna said the first eight volunteers with available results had developed neutralizing antibodies at levels similar to those in recovering patients. The hope is that these antibodies might block future infection. While these findings are encouraging, much remains unknown. Researchers at Oxford testing a different vaccine say it could be ready for emergency use as soon as this fall.

This timeline aligns with President Trump’s promise of a vaccine by year’s end. The head of the White House’s vaccine acceleration program, Operation Warp Speed, has suggested that “a few hundred million doses” could be delivered by then. These claims have been criticized as unrealistic; they may be an election-year strategy, encouraging perceptions that the administration is winning the battle against the coronavirus.

Already, the White House has been accused of pressuring government agencies to put politics before science when it comes to drugs to treat covid-19. Pressure to move quickly on a vaccine will be intense, and at some point the Food and Drug Administration will be asked to authorize a vaccine candidate for widespread use. Although the FDA could allow use of a coronavirus vaccine based on weak evidence, it must not make that mistake.

Since 2004, the FDA has had special authority to issue temporary authorizations for use of medical products when a public health emergency involves a biological agent capable of causing a serious or life-threatening disease and the totality of available evidence suggests that the product “may be effective” for prevention, diagnosis or treatment. The agency also must judge that the intervention’s known and potential benefits outweigh its known and potential risks, and that there is “no adequate, approved, and available alternative.” This standard is considerably weaker than that required for FDA marketing approval.

So far, the FDA has issued more than 150 emergency authorizations for covid-19, more than the total number issued before the pandemic. Mostly, these authorizations have been for tests, but they’ve also included personal protective equipment, ventilators and drugs such as hydroxychloroquine, chloroquine and remdesivir.

The potential benefits of emergency authorization for a therapeutic drug are clear: seriously ill patients get access to a therapy that might help them. But there are downsides. The drug might be unsafe or ineffective. Availability of the drug also might divert patients from trying more promising alternatives or participating in clinical trials needed to demonstrate what actually works.

The trade-offs are very different when considering emergency authorization for products intended to prevent rather than treat disease. The FDA has used this pathway for a vaccine only once, in 2005, allowing a decades-old vaccine to be used to prevent anthrax in a bioweapon attack. When the agency authorized emergency use of other preventive drugs, as it did for anthrax in 2008 and H1N1 influenza in 2009, the drugs were already FDA-approved. Emergency authorization of a new vaccine for the novel coronavirus would be unprecedented.

Emergency treatments may be given to thousands, or tens of thousands, of very sick patients. But a vaccine could reach millions — even tens of millions or hundreds of millions — of healthy people. Because vaccine recipients are not sick, their potential to benefit stands to be much less than that of seriously ill patients given a drug, especially in light of the effectiveness of alternatives such as physical distancing and good hygiene. If data ultimately shows that vaccine effectiveness is limited while the risks are substantial, harms to millions of uninfected individuals could be enormous. False reassurance from a flawed vaccine might also worsen the epidemic, setting back progress.

Public trust in vaccines is another critical consideration. Already, the anti-vaccine movement has begun to sow doubt about a coronavirus vaccine. Even if an option ultimately proves to be safe and effective, it won’t do much good if many people refuse it. And if a covid-19 vaccine causes harm, efforts to counter other vaccine-preventable diseases may be set back as well.

For all these reasons, the FDA must apply rigorous standards when considering whether to authorize a coronavirus vaccine. The agency must wait until data from a well-controlled trial demonstrates that a vaccine is able to prevent covid-19 and that it is safe enough to give to millions of people.

There is acute need for a vaccine that will change the course of this pandemic and allow normal economic and social activity to resume. Experience suggests that the FDA will be under intense pressure to authorize emergency use of the first plausible candidate, even if data supporting its benefits are thin. But when it comes to vaccines, “may be effective” is not good enough.

Read more: